Key ideas
Core concepts
- Reliable data about learning comes from seeing responses from every student, not just confident volunteers.
- “I don’t know” is not an acceptable end point: every student must engage with the thinking.
- Teach explicitly first, then check. Questioning is for assessing understanding, not for delivering content by “guess what’s in my head”.
- Withhold hints until after the check, so the data reflects what students can actually do.
Check for understanding is the systematic gathering of reliable data about student learning through strategic questioning and assessment techniques, enabling responsive teaching decisions (Black & Wiliam, 1998; Hattie, 2009).
Connected to
Responsive Teaching | Mini-Whiteboards | Cold-Call | Culture of Error | No Opt-Out | Diagnostic Questions
The problem it solves
Many teachers spend excessive time explaining concepts students already understand whilst rushing through areas of genuine confusion, because they lack reliable data about student learning. Systematic checking provides data to allocate time based on actual student needs, identifies misconceptions before they become entrenched, and lets teachers adjust instruction on evidence rather than intuition. It also ensures most students are progressing, not just confident volunteers (Black & Wiliam, 1998; Rosenshine, 2012; Hattie, 2009).
Basing understanding on responses from confident volunteers creates an illusion of class-wide comprehension. Volunteers provide an unrepresentative sample. Most students can mentally opt out whilst appearing engaged, the same students dominate participation, and the teacher forms a false sense of class progress from their strongest learners. The teacher moves on, unaware of growing gaps.
Several methods sample the whole class instead (Lemov, 2015; Rosenshine, 2012). Mini-whiteboards show every response immediately for any question type, making them the primary checking tool. Diagnostic questions work efficiently whilst revealing misconceptions during concept checking. Strategic cold-call enables targeted sampling across ability levels for follow-up questions, and multiple cold calls widen the sample when mini-whiteboards are unavailable.
Getting honest data
Allowing students to opt out with “I don’t know” sets the expectation that participation is optional. Every student must engage with thinking through questions like “What do you know about this?”, “What’s your best guess?”, or “Listen to Sarah’s answer, then repeat it back and tell me if you agree.” The message: in this classroom, thinking is non-negotiable, but being wrong is completely acceptable.
Pre-empting student difficulties with helpful hints feels supportive but undermines assessment validity. Hints like “Watch out, there are numbers you don’t need here,” “Be careful with this one (watch for hidden brackets),” or “Remember, we need to add the powers” create several problems. The teacher will never know if students actually needed the hint; they might have succeeded independently. The assessment data becomes invalid because success reflects the hint, not understanding. The cognitive demand of the question drops, and students learn to wait for teacher support rather than thinking independently. Do the check first, then provide hints based on what the check reveals, letting evidence guide support rather than assumptions.
Teachers also accidentally reveal whether answers are correct: verbal tells (“Are you sure?” only for wrong answers), physical tells (smiling for right answers, sighing for wrong ones), and pattern tells (always asking correct answers first). Use consistent stock phrases (“Thank you” regardless of correctness), ask students whether you have obvious tells, vary the order of questioning, and hold off confirming correctness until after discussion.
Students copying or hearing answers before they have thought independently undermines the check as well. Front-load the participation method (tell students how to respond before asking the question), have mini-whiteboards hover face down to prevent a copying cascade, use “heads down” voting for ABCD cards or fingers, set clear expectations (“I want your thinking, not your neighbour’s”), and address calling out or copying immediately.
Questioning technique
Rather than asking “Can anyone tell me…?”, choose who responds. Ask the question first, then name a student (“What’s the first step?… Sarah”), which keeps everyone thinking. Allow adequate wait time (minimum 3 seconds) for thought to develop. Question the students least likely to know, since targeting struggling students first gives more honest assessment. Require full sentences (“Give me a complete answer”) to deepen processing.
For the most honest picture, target students who struggled with the topic before, those who missed previous lessons, lower prior-attaining students, students suspected of not listening, and those with identified misconceptions. If they understand, it is reasonable to assume others do too, which makes this more reliable than asking confident volunteers. Balance is needed: mix in other students to avoid an obvious pattern.
Avoid using questioning to deliver content that should be taught explicitly (“guess what’s in my head”). The better sequence is to teach clearly, check whether students can retrieve or apply the content, then respond to what the check shows. Asking “How do you think we solve this equation?” before teaching is poor practice. Teach the method explicitly, then ask “What’s the first step to solve this equation?”
When students get questions wrong, the cause could be conceptual confusion or simple inattention during the explanation. Regular checks for listening separate the two: “What did I just say was the first step?”, “What was the last instruction I gave?”, “What did Sarah just explain?” These rule out attention issues, sustain focus during explanations, and allow a targeted response to genuine confusion.
Diagnostic questions
Diagnostic questions are multiple-choice questions with one correct answer and three wrong answers (distractors) designed to reveal specific misconceptions. Teachers can see all student responses quickly, wrong answers are interpretable because they map to specific misconceptions, and specific concepts can be assessed efficiently. They are more cognitively challenging than they appear. Run them with mini-whiteboards or technology, plan 5-10 seconds of thinking time plus 10 seconds of rehearsal, always clarify the correct answer afterward, and use wrong answers as teaching opportunities.
Responding to answers
Accepting partially correct answers and adding the missing details yourself (“rounding up”) creates problems. Students think they understand when they don’t, standards fall because partial answers become acceptable, the teacher does the cognitive work students should do, and students lose the chance to recognise their own gaps. Push for excellence instead: “Let’s improve this answer together,” “What’s the proper term for ‘cancelling’?”, “Can you be more precise with your language?”, “What do you mean by that?”
After an initial response, stretch it with follow-up questions: “How do you know?”, “Can you explain your reasoning?”, “What made you think of that approach?”, “How is this similar to yesterday’s problem?” This tests depth against surface understanding, requires students to articulate reasoning, reveals whether correct answers are correct for the right reasons, and builds mathematical communication skills.
Whole-lesson checks
Exit tickets use 2-3 specific questions about the day’s learning at the end of the lesson as a quick temperature check, informing planning for the next lesson and identifying students needing additional support. The “divide, dig, decide” process sorts responses into three categories: “yes” (students definitely got it), “maybe” (students partially understood or included elements of a good answer), and “no” (students definitely did not get it).
Rather than asking “Any questions about the homework?”, have students write questions anonymously on the board as they enter. This removes personal risk from asking questions, often reveals more widespread confusion, makes students more honest about what they don’t understand, and provides better data about common struggles.
When checking goes wrong
If students still hide confusion, examine the culture of error (is it genuinely safe to be wrong?), use more anonymous systems, model your own uncertainty and mistakes, and celebrate students who ask questions or admit confusion.
If responses don’t represent true understanding, reduce hints and leading questions, increase use of mini-whiteboards over volunteers, use diagnostic questions with good distractors, and follow up correct answers with “How do you know?”
If checking takes too much time, plan follow-up questions in advance, use efficient systems like ABCD cards, focus checks on the most critical concepts, and build routines so procedures become automatic. If students aren’t engaging, apply no opt-out consistently, check that questions are appropriately challenging, use warm call to build confidence, and keep the success rate around 80%.
References
Black, P., & Wiliam, D. (1998). Assessment and classroom learning. Assessment in Education: Principles, Policy & Practice, 5(1), 7-74. https://doi.org/10.1080/0969595980050102
Hattie, J. (2009). Visible learning: A synthesis of over 800 meta-analyses relating to achievement. Routledge.
Lemov, D. (2015). Teach like a champion 2.0: 62 techniques that put students on the path to college. Jossey-Bass.
Rosenshine, B. (2012). Principles of instruction: Research-based strategies that all teachers should know. American Educator, 36(1), 12-19, 39.